Académique Documents
Professionnel Documents
Culture Documents
00
Copyright0 1993PergamonPressLtd
CROSS-CULTURAL
ADAPTATION OF HEALTH-RELATED
QUALITY OF LIFE MEASURES: LITERATURE REVIEW
AND PROPOSED GUIDELINES
FRANCIS GUILLEMIN,* CLAIRE BOMBARDIER*~~
and DORCAS BEATON~
School of Public Health, University of Nancy, France, *ClinicalEpidemiology Division, Wellesley
Hospital, University of Toronto and fOntario Worker Compensation Institute, Toronto, Canada
(Received in revised form
11 June
1993)
Abstract-Clinicians
and researchers without a suitable health-related quality of life
(HRQOL) measure in their own language have two choices: (1) to develop a new
measure, or (2) to modify a measure previously validated in another language, known
as a cross-cultural adaptation process. We propose a set of standardized guidelines for
this process based on previous research in psychology and sociology and on published
methodological frameworks. These guidelines include recommendations for obtaining
semantic, idiomatic, experiential and conceptual equivalence in translation by using
back-translation techniques and committee review, pre-testing techniques and re-examining the weights of scores. We applied these guidelines to 17 cross-cultural adaptation
of HRQOL measures identified through a comprehensive literature review. The
reporting standards varied across studies but agreement between raters in their ratings
of the studies was substantial to almost perfect (weighted K = 0.66-0.93) suggesting that
the guidelines are easy to apply. Further research is necessary in order to delineate
essential versus optional steps in the adaptation process.
Health
Quality of life
Validity
Guidelines
status
Cross-cultural
index
RATIONALE
A large
comparison
Culture
1417
1418
FRANCIS
GUILLEMIN
et al.
METHODS
Development of guidelines
!I\i
/Exct
criteria
L
188 refs
Excerpta Medica
(1990-92)
6 abstracts
) Rejection
I
32 references
uin
L
Title and abstract review
Reference ma!,
= 712 references J
4
273 refs
5
573 refs
~c-3
Health Planning
and Administration
(1975-92)
Medlinc
(1966-92)
1420
FRANCISGUILLIMN
et al,
Our literature review identified several settings for the cross-cultural adaptation process.
A range of situations may be encountered depending on similarities and differences between
the cultures and languages of the populations
concerned. An instrument originally developed
in the English language in the U.S.A. can
readily be used by a majority of the American
population (Table 2: example 1).
Immigrants using the same language may
encounter particular problems in expressing
themselves with regard to health and HRQOL.
Therefore, particular attention should be paid
to the adaptation of cross-cultural HRQOL
measures to such populations. Immigrants to
the U.S.A., for instance Hispanics, will judge
their health and related QOL according to their
cultural origin and their degree of assimilation
into the host culture. They may have been
settled for long enough to have mastered the
English language sufficiently well to answer the
original instrument, and still refer to their
Spanish culture in assessing their situation
(Table 2: example 2).
An instrument used in a country other than
that in which it was developed may require
adaptation if the populations concerned have
another culture with similar language. For instance British English should be used in Great
Britain rather than American English. There are
sufficiently meaningful differences between the
British and American cultures to necessitate
modification of some items and validation of the
measure in its new setting [37-391 (Table 2:
example 3).
Recently settled immigrants with a low degree
of acculturation may require an instrument that
is cross-culturally adapted to their Spanish
(native) language and culture, but appropriate
to the American situation (Table 2: example 4).
Under most circumstances, instruments require adaptation for use in a different country
with both a different culture and a different
language. For instance, the American measure
would need to be modified for use in the French
language in France or in Canada (Table 2:
1421
1422
FRANCISGUILLWN et al.
Table 3. Guidelines to preserve equivalence in cross-cultural adaptation of HRQOL measures.*
1. Translation
Produce several translations
Use qualified translators
Back-translation
as translations
Committee review
Check for equivalence in source and final versions using a pre-test technique
Either use a probe technique
Or submit the source and 6nal versions to bilingual lay people
Immigrants: Choose the language of administration or use a dual-format measure
Weighting of scores
2. Back-translation
Produce as many back-translations as translations. Back-translation, translating back from
the final language into the source language, has
been shown to help improve the quality of the
final version [7,14]. Each first translations
should be back-translated independently from
each other. Misunderstandings in the first translation may be amplified in the back-translation,
and thereby revealed. Failure to adapt to the
cultural target context and ambiguity in the
source version can also be uncovered.
Use appropriate back-translators.
Backtranslation is of better quality if those who do
it are fluent in the idioms and colloquial forms
of the source language, i.e. the result of their
back-translation. Thus, they should also translate into their mother tongue. Unlike some of
the fist translators, back-translators
should
preferably not be aware of the intent and concepts underlying the material. Back-translators
without a priori knowledge of the intent of the
original instrument are free of biases and expectations and their back-translation may reveal
unexpected meanings or interpretations in the
final version.
3. Committee review
Constitute a committee to compare source and
jinal versions. A committee should be constituted in order to produce a final version of the
modified measure based on the various trans-
lations and back-translations obtained as described above. Part of that committees role
should also review the introduction and instruction to the questionnaire as well as review the
scaling of responses to each question (i.e. the
translation should maintain equivalence of steps
in Likert-type scales).
Membership in the committee should be multidisciplinary. To use the analogy of the development of a new health status measure, the
committee should consist of individuals expert
in the disease(s) explored, and in the intent of
the measure and the concepts to be explored.
Bilingual members are of particular value for
such committee [ 151.
In case of a cross-cultural adaptation for an
immigrant population, individuals representative of the target group are likely to be available.
Their input is likely to result in a measure better
adapted in terms of idioms and colloquialisms
than that which would be produced by highly
educated people [9,42]. A scale referring to the
ability to speak, write, read and understand
both languages has been developed [16] and can
be useful in selecting these bilingual committee
members.
Use structured techniques to resolve discrepancies. The committee may resolve problems by
considering the material it has now collected. It
may further decide to repeat the translation-back-translation
process. A decentring
technique [17] has also been proposed as a way
@semantic equivalence
1423
is equivalence in the
meaning of words, and achieving it may present problems with vocabulary and grammar.
For example, vocabulary problems may be
encountered in the question are you able to
bend ? which can refer to several parts of the
body, such as the arm, back or knees, and
might have been intended-and
translatedonly to explore the ability to flex (arm), bend
over (back) or squat (knees). Furthermore,
some words, such as happy, have several
subtly different meanings depending on the
context.
Grammatical
alterations are sometimes
necessary in the construction of sentences. For
example, languages without the gerund form
may be more difficult to adapt [13]: activities
couched in terms such as dancing, singing or
Modifv instructions orformat, modifv or reject
eating
(gerund form of to dance, sing and eat)
inappropriate items, generate new items. The
may
not
be readily translatable.
committee must ensure that the introduction to
??idiomatic equivalence. Since idioms and collothe research tool and the instructions for filling
quialisms are rarely translatable, equivalent
in the questionnaire are carefully translated in
expressions have to be found or items have to
order to preserve the replicability of the measure
be substituted. This is more likely to be necess[19]. The redundancy principle, i.e. repeating the
ary in the emotional and social dimensions.
same instruction in a different manner, may help
For example, Do you feel downhearted and
to reduce comprehension errors [12].
blue? or Do you feel at home? are untransThe review committee is also likely to modify
latable idioms for which equivalents must be
or eliminate irrelevant, inadequate and ambigufound. The item I am feeling on edge in the
ous items and may generate substitutes better
NHP was translated into I have my nerves
fitting the cultural target situation while mainoutside my skin in Italian [9], I feel nervous,
taining the general concept of the deleted items.
tense in French [43] and I am afraid in
Ensure that the translation is fully comprehenArabic [60].
sible. Guidelines about how to produce trans??experiential equivalence. The situations evoked
lations comprehensible to a majority of people
or depicted in the source version should fit the
have suggested using language which can be
target cultural context. This may result in the
understood by 10 to 1Zyear old children [12].
modification of an item. For example, in the
Recommendations include: short sentences with
Brazilian version of the HAQ, using public
key words in each item as simple as possible [ 111;
transportations was substituted for using a
the active rather than the passive voice; repeated
private car, since most of the people in Brazil
nouns instead of pronouns; and specific rather
have no car [44]. I have forgotten what it is
than general terms. Authors should avoid using:
like to enjoy myself [60} and How many
metaphors and colloquialisms; the subjective
hours a week do you have leisure activities?
mode; adverbs and prepositions telling where
do not refer to usual experiences in a number
and when, possessive forms; words indicating
of cultures and equivalent feelings (enjoy) or
vagueness; and sentences containing two differactivities (leisure) must be found or the items
ent verbs that suggest different actions.
discarded.
Verifv cross-cultural equivalence of source and ??conceptual equivalence refers to the validity of
final versions. Several taxonomies of cross-culthe concept explored and the events experitural equivalence have been proposed in the
enced by people in the target culture, since
psychiatry literature [12, 13,60,61]. The ultiitems might be equivalent in semantic meaning
mate parity is equivalence of HRQOL concepts
but not conceptually equivalent.
within the cultures concerned. Translators aimFor example, cousin and brother may
ing for conceptual equivalence should consider
mean more than simply second or first-degree
the following:
relative of the same generation. In many culof improving cross-cultural adaptation. This
technique considers the source and final versions equally important. Both are open to
modification during the translation procedure.
In other words, the measure is not considered to
centre on one of the languages. Decentring is
best conducted in close collaboration with the
authors. If problem items are found, the authors
may provide a working version of the instrument or items, maintaining the concept of the
questions, but avoiding colloquialisms. Searching for a common way to express a concept in
both languages is the best way to ensure that the
final version maintains content validity. It is
unusual for authors to be available, and this
process may need to be conducted by committee
members.
1424
FRANCIS
GUILLEMIN
et al.
5. Weighting scores
Consider adapting the weights of scores to the
cultural context. A scoring method using
weights is provided with the source versions of
some instruments (Sickness Impact Profile [64],
NHP [65]) in order to combine the information
in an index or in several indices (profile). However, the weighting may not apply to the new
cultural situation. It can be reexamined either
by judgement or using a mathematical approach. Using judgement, the cross-cultural
validity of the weighting of items is reexamined
by experts, who may be health care professionals, patients or lay people. Several techniques are available to elicit culture-adapted
weights from expert opinion. With a mathematical approach, data obtained from a
sample of patients are analysed by various
statistical techniques for scalability (Gutmann
analysis) or dimensionality (factor analysis) in
order to work out the best way of aggregating
the information in one index or several indices.
1425
Language
Sweden
France
Brazil
Canada
U.K.
3 translations
(2 translators in
each team, aware
of objectives and
disease)
1 translation by 2
translators aware
of objectives and
intent
1 translation
1 translator aware
of objective and
disease
-
N/A
N/A
U.K.
N/A*
Translation
U.K.
Country
3 back-translations
(1 English-cultured)
1 back-translation
by 2 translators
N/A
N/A
N/A
Back-translation
2 Rheumatologists
2 English professors
2 Physiotherapist
1 Occupational
therapist
5 RA patients
Authors + 2 bilinguals
2 bilingual
researchers
Committee
64 RA patients
Check relevance
31 RA patients
with translation
22 RA patients
with substitute version
Probe technique
30 RA patients
Probe technique:
questionnaire
+ interview
27 RA patients
Check understandability of the original: 22% pbs
33 RA patients
Check understability of modified
HAQ: 6% pbs
-
Pre-testing
Table 4. Reviewed articles with description of the methodology used for the cross-cultural adaptation of HRQGL measures
Griginal
instrument
1426
N/A
WA
N/A
135 consumers
(stratified by age/sex)
54 health profi 46
nurses + 8 physicians
Thurstone method
N/A
N/A
Weighting score
France
Spain
U.K.
U.S.A.
U.S.A.
U.S.A.
U.S.A.
U.S.A.
Sweden
Sweden
2 tranlations by 3
translators each: 1
literal + 1 idiomatic
2 translations
(1 translator each)
several translations
4 translations: 3 by
1 translator and 1
by 2 translators
2 translations: 1
idiomatic + 1 literal (3 translators
each)
3 previous translations
1 translation (1)
2 reviewers for 2
previous + 2 new
translations
3 authors
1 back-translation
(1 translator)
bilinguals
3 English and 3
1 back-translation
by other bilinguals?
1 backtranslation (1)
4 back-translations, each by 4
bilingual translators
(1)
2 previous backtranslation
2 back-translations
N/A
N/A
N/A
12 selected bilinguals
Check comprehension and
readability
Cardiology
+ rheumatology
patients
Check acceptability
2 experts
1 back-translation
undetailed
2 translations
(1 translator) by
experts in NHP
10 unskilled
workers
Bilinguals
1 back-translation
1 translation
N/A
In a pilot study
1 back-translation
by 1 translator
1 translation by 2
translators aware
of objective and
disease (authors)
-
ii
B
=
c1
i:
I
1428
FR+NC~SGUILLEMINet al.
Number of
studies
Individual guidelines (5 sections)
Translation
Back-translation
committee
Pre-testing
Weighting scores
Overall guidelines
Similar language, other country
Other language, other country
Other language, same country
(immigrants)
All articles
12
10
10
9
6
Intra-class
correlation
coefficient
2.1 [l-3]
1.9 [l-3]
2.2 [l-3]
2.1 [l-3]
2.4 [2-31
Weighted kappa
0.70
0.66
0.86
0.88
0.93
3
8
6
0.8 [0.5-l]
1.3 [O&2.1]
1.6 [OA-2.61
1
0.96
0.87
17
1.3 [O/l-2.6]
0.92
*Mean score across studies calculated as the mean quality ratings assigned for each section with the following values:
Individual guidelines: good = 3, moderate = 2, poor = 1;
Overall guidelines: good = 3, moderate = 2, poor = 1, 0 = not done.
1429
provides a common measure for the investigation of HRQOL within different cultural
contexts;
??it offers a standard measure for use in international studies, many of which are now being
conducted;
??it
allows comparisons between national/
cultural groups relying on a standard measure
designed and adapted to measure the phenomenon cross-culturally;
??it allows the inclusion of immigrants avoiding
the frequent bias of representing only the
dominant culture of the country;
??it is less costly and time-consuming than generating a new measure. Nevertheless, it should
be borne in mind that the cross-cultural adaptation of HRQOL also requires careful attention, involves numerous people and is
time-consuming.
Acknowledgements-The authors are grateful to Dr Stephen
Stansfeld and Dr James Wright for fruitful comments on
earlier drafts of this manuscript. F. Guillemin was supported by a Bourse Lavoisier from the French Ministry of
Foreign Affairs and by the Fondation pour la Recherche
Therapeutique.
REFERENCES
1. Spilker B, Simpson RL, Tilson HH. Quality of life
bibliography and indexes: 1991 update. J Clin Res
WarntaeoePidemlol 1992; 6: 205-266.
2. Aaronson NK, Meverowitx BE. Bard M et al. Oualitv
of life research in~oncology. Past achievements and
future priorities. Caneer 1991; 67: 839-843.
3. Fitzpatrick R, Fletcher A, Gore S, Jones D, Spiegelhalter D, Cox D. Quality of life measures in health
care. I: Applications and issues in assessment. Br Med
J 1992; 305: 10741077.
4. Advances in Health Status Assessment: Proceedings of
a conference. Med Care 1992; 30: MSl-MS293.
5. Gerin P, Daxord A, Cialdella P, Leizorovicz A, Boissel
JP. Le questionnaire Protll de la qualite de la vie
subjective. Premiers elements de validation. Tberapie
1991; 46: 131-138.
6. Kempen GI, Suurmeijer TP. The development of a
hierarchical polychotomous ADLIADL
scale for
1430
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
FRANCIS
GUILLEMIN
et of.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
46.
41.
48.
49.
50.
51.
52.
53.
54.
55.
56.
51.
58.
1431
(Appendix overleaf)
FRANCIS
GUILLEMIN
et al.
1432
APPENDIX
ARSTRACIION FORM
FOR THE APPRAISAL OF CROSS-CULTURAL ADAPTATION OF HRQOL INSTRUMENTS
Originalinstrument:
Population addressed by the target version:
Language:
culture:
Country of origin:
Country of residence:
SCORE
1. Tradaiion
tedoliqoe:
Number of translations:
Number of translators in each translation:
Yes
Yes
Yes
No
No
No
Yes
No
Yes No
2. Back-tramlatilIn technique
Number of back-translations:
Number of translators in each back-translation:
Yes
No
3. committee approach
Yes No
Yes
No
With bilinguals:
Yes
No
With monolinguals:
using a probe technique:
using another technique:
Yes
Yes
No
Yes
No
Translation/back-translation
process iterated:
??
4. Pre-teat@
No
specify:
Sample composition:
??
Sample size:
5. woighthg
SCORE
seorea adoptatioo
Yes
Yes
Yes
No
No
No
Not applicable
??